newsdiscourse-model

This model is a fine-tuned version of roberta-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.9458
• F1: 0.5610

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 1e-05
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10.0

Training results

Training Loss	Epoch	Step	Validation Loss	F1
No log	0.14	100	1.4843	0.2881
No log	0.28	200	1.3307	0.3841
No log	0.43	300	1.2427	0.3991
No log	0.57	400	1.2590	0.4899
1.2399	0.71	500	1.2648	0.4658
1.2399	0.85	600	1.2064	0.4988
1.2399	1.0	700	1.2564	0.4668
1.2399	1.14	800	1.2062	0.4912
1.2399	1.28	900	1.1202	0.4904
0.9315	1.42	1000	1.1924	0.5188
0.9315	1.57	1100	1.1627	0.5034
0.9315	1.71	1200	1.1093	0.5111
0.9315	1.85	1300	1.1332	0.5166
0.9315	1.99	1400	1.1558	0.5285
0.8604	2.14	1500	1.2531	0.5122
0.8604	2.28	1600	1.2830	0.5414
0.8604	2.42	1700	1.2550	0.5335
0.8604	2.56	1800	1.1928	0.5120
0.8604	2.71	1900	1.2441	0.5308
0.7406	2.85	2000	1.2791	0.5400
0.7406	2.99	2100	1.2354	0.5485
0.7406	3.13	2200	1.3047	0.5258
0.7406	3.28	2300	1.3636	0.5640
0.7406	3.42	2400	1.2963	0.5747
0.6355	3.56	2500	1.2897	0.5123
0.6355	3.7	2600	1.3225	0.5481
0.6355	3.85	2700	1.3197	0.5467
0.6355	3.99	2800	1.2346	0.5353
0.6355	4.13	2900	1.3397	0.5629
0.5698	4.27	3000	1.4259	0.5622
0.5698	4.42	3100	1.3702	0.5607
0.5698	4.56	3200	1.4294	0.5584
0.5698	4.7	3300	1.5041	0.5459
0.5698	4.84	3400	1.4156	0.5394
0.5069	4.99	3500	1.4384	0.5527
0.5069	5.13	3600	1.5322	0.5439
0.5069	5.27	3700	1.4899	0.5557
0.5069	5.41	3800	1.4526	0.5391
0.5069	5.56	3900	1.5027	0.5607
0.4127	5.7	4000	1.5458	0.5662
0.4127	5.84	4100	1.5080	0.5537
0.4127	5.98	4200	1.5936	0.5483
0.4127	6.13	4300	1.7079	0.5401
0.4127	6.27	4400	1.5939	0.5521
0.3574	6.41	4500	1.5588	0.5702
0.3574	6.55	4600	1.6363	0.5568
0.3574	6.7	4700	1.6629	0.5535
0.3574	6.84	4800	1.6523	0.5662
0.3574	6.98	4900	1.7245	0.5461
0.3417	7.12	5000	1.6766	0.5629
0.3417	7.26	5100	1.8219	0.5450
0.3417	7.41	5200	1.7422	0.5533
0.3417	7.55	5300	1.8250	0.5564
0.3417	7.69	5400	1.7744	0.5600
0.2852	7.83	5500	1.7919	0.5549
0.2852	7.98	5600	1.7604	0.5639
0.2852	8.12	5700	1.7660	0.5599
0.2852	8.26	5800	1.7323	0.5600
0.2852	8.4	5900	1.9174	0.5529
0.2606	8.55	6000	1.8664	0.5611
0.2606	8.69	6100	1.9191	0.5568
0.2606	8.83	6200	1.8900	0.5565
0.2606	8.97	6300	1.9376	0.5524
0.2606	9.12	6400	1.9220	0.5594
0.2274	9.26	6500	1.9188	0.5585
0.2274	9.4	6600	1.9459	0.5527
0.2274	9.54	6700	1.9439	0.5543
0.2274	9.69	6800	1.9437	0.5596
0.2274	9.83	6900	1.9484	0.5581
0.2258	9.97	7000	1.9458	0.5610

Framework versions

• Transformers 4.30.2
• Pytorch 2.0.1+cu117
• Datasets 2.13.1
• Tokenizers 0.13.3